ICCL Files Complaint Against European Commission Over Generative AI Usage in Public Documents

ICCL Takes Bold Step Against EU Commission’s Use of Generative AI

The Irish Council for Civil Liberties (ICCL) has lodged a formal complaint with the European Ombudsman, targeting the European Commission’s controversial use of generative AI in public documents. This action raises serious questions about adherence to both internal guidelines and overarching treaty obligations aimed at ensuring transparency and accuracy in governmental communications.

A Revelation from the European Commission

A recent document request from ICCL unveiled an unexpected dimension to the European Commission’s operations. Among routine links provided in their response, one stood out: it contained a traceable marker, “utm_source=chatgpt.com,” indicating that at least one of the links was generated using OpenAI’s ChatGPT. This revelation not only suggests that the Commission might be basing public communications on generative AI outputs but also raises concerns about consistency and reliability.

Given that generative AI can produce information that is sometimes correct and sometimes incorrect— errors in these systems are not mere glitches but part of their fundamental design—the implications for public information are significant. Relying on such tools can lead to misinformation, which contradicts the EU’s commitment to providing accurate information to its citizens.

The Legal and Ethical Implications

EU institutions have an implicit duty to ensure that their communications reflect accurate and trustworthy information. By potentially utilizing AI-generated content, the Commission risks violating its treaty obligations to uphold standards of good administration. As the ICCL points out, this reliance on so-called “bullshit generators” is not just a lapse in practice; it could represent a significant legal misstep.

Moreover, ICCL argues that using generative AI likely contravenes the Commission’s own staff guidelines, which explicitly prohibit the replication of AI-generated outputs in public documents. This inconsistency raises critical questions about the effectiveness of internal controls and oversight mechanisms.

Call for Transparency

Dr. Kris Shrishak, Senior Fellow at ICCL, has emphasized the need for transparency when public bodies engage generative AI tools. “Public bodies like the European Commission should always be transparent and disclose if a generative AI tool is used in any public document,” he stated. This transparency is not only a good practice; it is essential for maintaining public trust.

Dr. Shrishak further asserted that if generative AI outputs are utilized, the onus should be on the Commission to prove the accuracy and integrity of that information, rather than on the public or recipients to prove otherwise. Without this accountability, the potential for misinformation could undermine the very fabric of public administration.

Conclusion

As the ICCL formally seeks accountability from the European Commission, this situation serves as a reminder of the increasing complexities surrounding AI technologies in governance. The use of generative AI in public documentation is a double-edged sword: while it offers potential efficiencies, it also poses distinct risks that must be addressed with diligence and responsibility.

The path forward requires not only adherence to established guidelines but also a commitment to transparency and accountability, ensuring that the public can trust the communications from their governmental institutions.

Contact Information for Media Queries

For any media inquiries or to schedule interviews, please reach out to:

• Ruth McCourt: ruth.mccourt@iccl.ie
• Molly Kavanagh: molly.kavanagh@iccl.ie
• Phone: 087 415 7162